A kit that gives AI agents cognitive abilities.

Choose and implement the right visual design system for any app or website. Brief-inference read with variance/motion/density dials, 14 design systems with when-to-use guidance plus buy-vs-build routing, a variance engine for non-repeating directions, 122 anti-AI-slop rules, component patterns (double-bezel cards, gapless bento, magnetic hover, micro-UIs), AIDA page structure, audit-and-upgrade workflow, DESIGN.md spec generation, anti-drift conventions, and UX psychology laws with a dark-pattern guardrail.

Build, scale, audit, and govern a design system as an engineering product. Covers token architecture, governance models, contribution tiers, versioning, deprecation, drift detection, multi-brand orchestration, Figma-to-code pipelines, CI gates, and AI-agent readiness.

Review code changes for real bugs, security issues, and convention violations with a confidence filter that suppresses noise. Covers correctness, security, state/concurrency, error handling, and project-convention violations, plus an optional simplicity pass that hunts over-engineering (reinvented stdlib, dead abstractions, unneeded deps) with one-liner findings. Severity-ranked findings with file:line citations, not style nitpicks.

Design REST API contracts before implementation: resource modeling, route conventions, request/response shapes, error codes, pagination, idempotency, and versioning. Produces consistent, predictable API contracts that respect existing codebase conventions.

Adversarial security review of a codebase. Maps attack surface, checks OWASP Top 10 in priority order (access control, injection, auth, crypto, misconfig, SSRF, XSS, secrets), audits dependency CVEs, and produces severity-ranked findings with exploitability confirmation and remediation. Defensive analysis only.

Design and build Model Context Protocol servers that expose tools, resources, and prompts to AI agents. Discovery-first design phase, deployment decision matrix (remote HTTP / MCP app / MCPB bundled / stdio), tool design patterns, elicitation vs widget routing, .mcp.json plugin integration, MCPB manifest and bundling, local security rules, runnable TypeScript and Python examples, and inspector testing.

Design persistent memory systems for AI agents. Four-layer architecture (soul rules, learned preferences, working memory, knowledge base), memory write rules, a self-improvement loop that runs as a periodic review process, confidence levels, memory poisoning defenses, and multi-tenant scoping.

Distill a large teacher model into a small model that runs on modest hardware. Covers task scoping, training data generation, LoRA fine-tuning with unsloth, student model sizing by RAM, evaluation against the teacher, GGUF quantization, and deployment with Ollama or llama.cpp.

Build retrieval-augmented generation systems: document ingestion, structure-aware chunking, embeddings, vector store selection by scale, hybrid retrieval, grounded answer generation with citations, multi-tenant isolation, and evaluation of retrieval hit-rate vs answer faithfulness.

Design and write marketing and landing pages that convert without looking AI-generated. Narrative arc, hero variations, conversion copy rules, pricing page patterns, one-memorable-element discipline, and an anti-slop blocklist for generic landing page patterns.

Design data-dense dashboards, admin panels, and internal tools that answer questions rather than decorate. Audience archetypes and KPI selection process, question-first hierarchy, chart selection rules, data table discipline, widget states, filters, density, and anti-patterns for dashboard design.

Design mobile app UIs that respect platform conventions instead of shrunken web pages. Four layers: pre-work (MFRI risk index, design bible locking), platform and navigation (iOS HIG vs Material 3, deep links, predictive back), visual design (60/30/10 color, 8-pt grid, system fonts, OLED dark mode, haptics), and execution (states, Peak-End rule, user-stage personalization, performance patterns, anti-AI-tells, 33-point quality checklist).

Design, audit, and refine UI components from small controls to complex navigation. Create professional, distinctive, product-grounded concepts while preserving behavior, accessibility, and responsive quality.

Audit and implement technical and on-page SEO to professional standard. Meta tags, canonicals, robots and sitemap rules, JSON-LD structured data, Open Graph, Core Web Vitals, EEAT signals, and anti-slop content checks.

Audit Apple-platform apps against App Review Guidelines, App Store Connect requirements, privacy and business rules; diagnose rejections and prepare factual responses. Covers guideline mapping, 2.3.10 screenshots, 3.2 distribution, and current-source verification.

Audit Android apps against Google Play Developer Program Policies and Play Console requirements; diagnose rejections and enforcement, reconcile Data safety, permissions, billing, ads, target API, and prepare evidence-based appeals.

Draft professional-grade privacy policies, terms of service, cookie policies, refund policies, and acceptable use documents. Intake-driven drafting matched to actual data practices with jurisdiction-aware clauses (GDPR, CCPA, COPPA).

Author, audit, and improve Agent Skills and agent configs against the open SKILL.md standard. Covers pushy trigger descriptions, three-level progressive disclosure, execution-flow body conventions, agent frontmatter, evals.json test cases, benchmark metrics, and the validate-fix-rerun iteration loop.